On 2/4/19 9:07 AM, Jeff Law wrote: > On 2/4/19 6:15 AM, Richard Biener wrote: >> >> When I introduced tree-form bitmaps I forgot to think about GC. >> The following drops the chain_prev annotation to make the marker >> work for trees. I've also maked the obstack member GTY skip >> (and prevent bitmap_obstack from gengtype processing) because >> the obstack isn't used for GC allocated bitmaps. >> >> Bootstrap & regtest running on x86_64-unknown-linux-gnu. >> >> Richard. >> >> 2019-02-04 Richard Biener <rguent...@suse.de> >> >> PR middle-end/89150 >> * bitmap.h (struct bitmap_obstack): Do not mark GTY. >> (struct bitmap_element): Drop chain_prev so we properly recurse on >> the prev member, supporting tree views. >> (struct bitmap_head): GTY skip the obstack member. > Was there a particular failure mode you observed or was this discovered > by inspection. > > The reason I ask is my tester is showing occasional failures > bootstrapping hppa-linux-gnu with a segfault in ICF. > > I'd rather not have to dig into it if it can be avoided :-) > Bootstrapping hppa via qemu is something like 6 hours... And just to follow-up. I'm really starting to wonder if the hppa issue is a qemu bug of some kind.
The faulting instruction, ACAICT, shouldn't be faulting. It appears to be loading from a valid address, it's a byte load (so no alignment issues). Furthermore, I can instruction-step over it in gdb. Anyway, I think we can reasonably conclude the PA issue is unrelated to your change. jeff
